前言：

ansible的结果默认是输出到cli终端和日志里面的，用惯了saltsatck的returners数据回调后，也很是喜欢ansible也有，一开始不知道有这个功能，自己也简单实现了这样的功能。

我的实现方式是，在模块里面做一些输出的逻辑。当使用ansible runner api的时候，是在后面runner代码，最后加了一段往redis输出的逻辑。 这里实现数据的输出有些独特，但是只能是在模块和 api方面搞 。 如果是用playbook的话，按照我以前的思路的话，再继续改ansbile的源码。  这两天听沈灿说，ansible有个callback_plugins的功能，可以对于执行的状态做一些判断，比如，执行成功，执行失败，异步执行，异步执行失败，playbook开始，结束等等。

沈灿这货先写了关于ansible callbacks的文章，我看到了后，才知道有而一个东西。大家可以看看 。

http://www.shencan.net/index.php/2014/07/17/ansible-%E6%8F%92%E4%BB%B6%E4%B9%8Bcallback_plugins-%EF%BC%88%E5%AE%9E%E6%88%98%EF%BC%89/

我也不说复杂了，就简单说一个例子，把执行的结果，都推到redis里面，也可以暂存到sqlite数据库里面，只是这段代码我给屏蔽了，有兴趣的朋友再搞搞。对于redis里面的数据可以写一个页面展现下，专门记录错误的问题，成功的就pass掉。

原文：http://rfyiamcool.blog.51cto.com/1030776/1440624

#xiaorui.ccimport os
import time
import sqlite3
import redis
import jsondbname = '/tmp/setup.db'
TIME_FORMAT='%Y-%m-%d %H:%M:%S'try:con = sqlite3.connect(dbname)cur = con.cursor()
except:passdef log(host, data):#    if type(data) == dict:
#        invocation = data.pop('invocation', None)
#        if invocation.get('module_name', None) != 'setup':
#            return
#
#    facts = data.get('ansible_facts', None)
#
#    now = time.strftime(TIME_FORMAT, time.localtime())
#
#    try:
#        # `host` is a unique index
#        cur.execute("REPLACE INTO inventory (now, host, arch, dist, distvers, sys,kernel) VALUES(?,?,?,?,?,?,?);",
#        (
#            now,
#            facts.get('ansible_hostname', None),
#            facts.get('ansible_architecture', None),
#            facts.get('ansible_distribution', None),
#            facts.get('ansible_distribution_version', None),
#            facts.get('ansible_system', None),
#            facts.get('ansible_kernel', None)
#        ))
#        con.commit()
#    except:
#        pass
#
class CallbackModule(object):def runner_on_ok(self, host, res):r = redis.Redis(host='127.0.0.1', port=6379, db=0) r.set(host,str(res))f = open('/tmp/11','a')f.write(str(host))f.write(str(res))f.close()log(host, res)def runner_on_failed(self, host, res, ignore_errors=False):f = open('/tmp/11','a')f.write('\nbad\n')f.close()log(host, res)

还是可以接收所有的facts数据的。

原文：http://rfyiamcool.blog.51cto.com/1030776/1440624

原文：http://rfyiamcool.blog.51cto.com/1030776/1440624

虽然我上面的例子用了redis，sqlite数据库，其实我个人推荐用mongodb这样的文档数据库的。因为ansible主runner函数，给callbacks传递了一个叫res的变量，他本身就是一个dict对象，如果放到redis的hash，sqlite的各种字段，够你烦的了，如果直接mongo，那就简单了，直接insert ！ 欧了

这里在show一个邮件的callbacks代码，场景是，非常消耗时间的任务，当执行完成后，查看结果咋办？　　但是你也可以在终端继续看，既然咱们讲了callbacks_plugins,就可以把结果push到你的邮箱里面，当然只给你发错误的，有问题的。 下面的callback代码需要自己替换成自己用的邮箱、密码、smtp服务器。

#xiaorui.cc
原文：http://rfyiamcool.blog.51cto.com/1030776/1440624 import smtplibdef mail(subject='Ansible error mail', sender='<root>', to='root', cc=None, bcc=None, body=None):if not body:body = subjectsmtp = smtplib.SMTP('localhost')content = 'From: %s\n' % sendercontent += 'To: %s\n' % toif cc:content += 'Cc: %s\n' % cccontent += 'Subject: %s\n\n' % subjectcontent += bodyaddresses = to.split(',')if cc:addresses += cc.split(',')if bcc:addresses += bcc.split(',')for address in addresses:smtp.sendmail(sender, address, content)smtp.quit()class CallbackModule(object):"""This Ansible callback plugin mails errors to interested parties."""def runner_on_failed(self, host, res, ignore_errors=False):if ignore_errors:returnsender = '"Ansible: %s" <root>' % hostsubject = 'Failed: %(module_name)s %(module_args)s' % res['invocation']body = 'The following task failed for host ' + host + ':\n\n%(module_name)s %(module_args)s\n\n' % res['invocation']if 'stdout' in res.keys() and res['stdout']:subject = res['stdout'].strip('\r\n').split('\n')[-1]body += 'with the following output in standard output:\n\n' + res['stdout'] + '\n\n'if 'stderr' in res.keys() and res['stderr']:subject = res['stderr'].strip('\r\n').split('\n')[-1]body += 'with the following output in standard error:\n\n' + res['stderr'] + '\n\n'if 'msg' in res.keys() and res['msg']:subject = res['msg'].strip('\r\n').split('\n')[0]body += 'with the following message:\n\n' + res['msg'] + '\n\n'body += 'A complete dump of the error:\n\n' + str(res)mail(sender=sender, subject=subject, body=body)def runner_on_unreachable(self, host, res):sender = '"Ansible: %s" <root>' % hostif isinstance(res, basestring):subject = 'Unreachable: %s' % res.strip('\r\n').split('\n')[-1]body = 'An error occured for host ' + host + ' with the following message:\n\n' + reselse:subject = 'Unreachable: %s' % res['msg'].strip('\r\n').split('\n')[0]body = 'An error occured for host ' + host + ' with the following message:\n\n' + \res['msg'] + '\n\nA complete dump of the error:\n\n' + str(res)mail(sender=sender, subject=subject, body=body)def runner_on_async_failed(self, host, res, jid):sender = '"Ansible: %s" <root>' % hostif isinstance(res, basestring):subject = 'Async failure: %s' % res.strip('\r\n').split('\n')[-1]body = 'An error occured for host ' + host + ' with the following message:\n\n' + reselse:subject = 'Async failure: %s' % res['msg'].strip('\r\n').split('\n')[0]body = 'An error occured for host ' + host + ' with the following message:\n\n' + \res['msg'] + '\n\nA complete dump of the error:\n\n' + str(res)mail(sender=sender, subject=subject, body=body)

如果不想发邮件，又不想搞到数据库里面，怎么办？　那来点低端的。　　直接写入到文件里面。

官方给出一个例子，大家照着模板写就行了。

import os
import time
import jsonTIME_FORMAT="%b %d %Y %H:%M:%S"
MSG_FORMAT="%(now)s - %(category)s - %(data)s\n\n"if not os.path.exists("/var/log/ansible/hosts"):os.makedirs("/var/log/ansible/hosts")def log(host, category, data):if type(data) == dict:if 'verbose_override' in data:# avoid logging extraneous data from factsdata = 'omitted'else:data = data.copy()invocation = data.pop('invocation', None)data = json.dumps(data)if invocation is not None:data = json.dumps(invocation) + " => %s " % datapath = os.path.join("/var/log/ansible/hosts", host)now = time.strftime(TIME_FORMAT, time.localtime())fd = open(path, "a")fd.write(MSG_FORMAT % dict(now=now, category=category, data=data))fd.close()class CallbackModule(object):"""logs playbook results, per host, in /var/log/ansible/hosts"""def on_any(self, *args, **kwargs):passdef runner_on_failed(self, host, res, ignore_errors=False):log(host, 'FAILED', res)def runner_on_ok(self, host, res):log(host, 'OK', res)def runner_on_skipped(self, host, item=None):log(host, 'SKIPPED', '...')def runner_on_unreachable(self, host, res):log(host, 'UNREACHABLE', res)def runner_on_no_hosts(self):passdef runner_on_async_poll(self, host, res, jid, clock):passdef runner_on_async_ok(self, host, res, jid):passdef runner_on_async_failed(self, host, res, jid):log(host, 'ASYNC_FAILED', res)def playbook_on_start(self):passdef playbook_on_notify(self, host, handler):passdef playbook_on_no_hosts_matched(self):passdef playbook_on_no_hosts_remaining(self):passdef playbook_on_task_start(self, name, is_conditional):passdef playbook_on_vars_prompt(self, varname, private=True, prompt=None, encrypt=None, confirm=False, salt_size=None, salt=None, default=None):passdef playbook_on_setup(self):passdef playbook_on_import_for_host(self, host, imported_file):log(host, 'IMPORTED', imported_file)def playbook_on_not_import_for_host(self, host, missing_file):log(host, 'NOTIMPORTED', missing_file)def playbook_on_play_start(self, name):passdef playbook_on_stats(self, stats):pass

原文： http://rfyiamcool.blog.51cto.com/1030776/1440624

也可以把结果以webhooks钩子的方式，做些你想做的东西。

callbacks的各种状态还是很多的，每个函数的字眼还是很好理解的。

比如：

on_any  哪都有他 ！任何的状态他触发。

runner_on_failed 失败

runner_on_ok  成功

runner_on_unreachable 网络不可达

runner_on_no_hosts 没有主机

runner_on_async_poll 任务的异步执行

playbook_on_start  playbook执行的时候

等等。。。。  自己尝试吧 ！

class CallbackModule(object):def on_any(self, *args, **kwargs):passdef runner_on_failed(self, host, res, ignore_errors=False):log(host, 'FAILED', res)def runner_on_ok(self, host, res):log(host, 'OK', res)def runner_on_skipped(self, host, item=None):log(host, 'SKIPPED', '...')def runner_on_unreachable(self, host, res):log(host, 'UNREACHABLE', res)def runner_on_no_hosts(self):passdef runner_on_async_poll(self, host, res, jid, clock):passdef runner_on_async_ok(self, host, res, jid):passdef runner_on_async_failed(self, host, res, jid):log(host, 'ASYNC_FAILED', res)def playbook_on_start(self):passdef playbook_on_notify(self, host, handler):passdef playbook_on_no_hosts_matched(self):passdef playbook_on_no_hosts_remaining(self):passdef playbook_on_task_start(self, name, is_conditional):passdef playbook_on_vars_prompt(self, varname, private=True, prompt=None, encrypt=None, confirm=False, salt_size=None, salt=None, default=None):passdef playbook_on_setup(self):passdef playbook_on_import_for_host(self, host, imported_file):log(host, 'IMPORTED', imported_file)def playbook_on_not_import_for_host(self, host, missing_file):log(host, 'NOTIMPORTED', missing_file)def playbook_on_play_start(self, name):passdef playbook_on_stats(self, stats):pass

原文： http://rfyiamcool.blog.51cto.com/1030776/1440624

咱们可以简单看看ansible的callbacks源码。

规定了两个类，一个是供应ansible-playbook用的，还有一个是供应ansible，也就是cli。 根据各种的情况，调用不同的函数，首先会打到终端，再log日志，最后是自定义的callbacks的插件。

好了，就这样了 ！！！！